Operon‑S1 is a fast decision controller that separates a language‑model agent’s judgment from its execution, posing 13 closed questions at key decision points and using a lightweight decision model to answer them. In seven life‑science benchmarks, the local decision model handled judgments in 0.6 s on median with no measurable impact on agent performance, while the main language model’s involvement accounted for 8–35 % of task cost. The system records every decision and its outcome, allowing retraining to improve accuracy, and is already applied to wet‑lab protocol troubleshooting and germline panel design.
Large language model (LLM) agents now fix software, operate computers and serve customers, but their harnesses, the programs that run them, call a large LLM at every step, which makes long tasks slow...
JevAgentHarness is a new framework that benchmarks, deploys, and trains decision models to share control with large language models (LLMs) in agentic tasks. It introduces AgentDecisionBench to evaluate decision models from isolated questions to full agent control, revealing that accurate isolated decisions still miss many tasks and that naïve delegation can reduce success. By fitting confidence thresholds per choice and training a 4B decision model on states after agent mistakes, JevAgentHarness achieves high task success while cutting cost and speeding up execution, with results and resources made publicly available.
The paper introduces TRACE, a protocol designed to diagnose whether changes in verifier scores for large language model agents reflect actual changes in agent behavior or merely alterations in the evaluation process. TRACE works by applying targeted changes to evaluation components, running paired experiments, and re‑scoring unchanged trajectories to isolate the source of score variation. Experiments on synthetic tasks and public benchmark tasks demonstrate that seemingly significant score shifts can often be attributed to evaluation artifacts, while TRACE can also detect genuine behavioral changes.
arXiv:2610.02588v1 Announce Type: new Abstract: Agents are increasingly given open-ended research tasks: discovering an empirical law from self-designed experiments, improving a heuristic whose optim...
The paper investigates how language‑model judges can make version‑dependent errors when evaluating upgraded agents. Using 35 public coding‑agent submissions, two customer‑service agents, and over a thousand expert‑labeled trajectories, the authors show that fixed judges often reject task‑conditioned error invariance and can incorrectly approve failed patches, especially as agent capability increases. Paired audits of current outputs reduce interval width only marginally, and the study concludes that independent human patch review is still necessary.
arXiv:2610.18298v1 Announce Type: new Abstract: What binds large language models on auditable, hypothesis-driven decision tasks, where every candidate can be checked against the evidence? On the publ...
System1-Arena is a benchmark that evaluates which decisions can be handled by System‑1 models—those that output probabilities in a single forward pass—versus System‑2 models that generate reasoning steps. It contains 92,000 questions across 105 tasks from 268 public sources, grouped into text classification, applied decisions, evaluation and control, and agent/action, and also includes 138 closed‑loop environments such as web shopping, coding control, robotics, games, and driving. The study tests 30 models, showing that large System‑1 models can match or exceed fast closed LLMs in non‑agentic accuracy with better‑calibrated probabilities at lower latency, while reasoning LLMs are most accurate but far more expensive, and large open LLMs perform best in closed‑loop settings.
The paper introduces the concept of an agent’s "taste"—its ability to make effective long‑horizon decisions—and presents Taste‑Bench, a new benchmark that automatically generates decision‑fork questions from agent trajectories. Taste‑Bench evaluates models on choosing the best path without seeing future outcomes, revealing that top models answer only about 60% of questions correctly and that later‑appearing evidence makes forks harder. The authors also demonstrate that training a student model to mimic a teacher’s judgment improves decision quality and overall success on held‑out software engineering tasks.
The paper reports on training and calibration experiments for three open models on the ARC‑AGI‑2 benchmark, using offline runs on four L4 GPUs for up to 12 hours. For the NVARC Qwen3‑4B grid model, a curated corpus of 12,000 executable rules and 95,993 tasks was built, and fine‑tuning on 1,920 real task records raised the semi‑private score from 28.96 to 33.06, with no synthetic‑task checkpoint outperforming the real‑task model. For the Qwen3.8‑27B reasoning model, fine‑tuning on 1,036 of its own reasoning traces and a specialized inference procedure increased the score from 26.81 to 44.03, while the Qwen3.8‑Flash‑Next mixture‑of‑experts model achieved a best score of 56.67 after expert pruning, routing statistics adaptation, and an FP8 indexer cache.
The paper introduces a new evaluation method called "same-input rerun" to assess the consistency of clinical language‑model agents across repeated runs. By replaying 1,000 MedAgentBench tasks with identical inputs, the authors find that action‑level outputs—such as test orders, medication requests, and referrals—vary significantly, even when benchmark scores remain unchanged. The study demonstrates that current benchmarks, which typically evaluate only a single run per task, can miss substantial behavioral divergence.
arXiv:2607. 01470v1 Announce Type: new Abstract: Clinical protocol-execution tasks -- checking a lab value, applying a threshold, placing a correctly structured FHIR order -- are natural candidates for RL from world feedback: once clinical SMEs encode decision logic into a verifier, that verifier grades unlimited rollouts without per-episode annotation.
The paper argues that evaluating agents as fixed models is insufficient, proposing instead to treat them as configurable systems. Using a new benchmark of four scientific tasks, the authors analyze how five configuration aspects—task information, reasoning, self‑verification, time budget, and backbone model—affect performance, noting that about 54% of outcome variance arises from run‑to‑run differences even with the same settings. The study finds that providing more task information has the strongest impact, while interactions among settings (e.g., extra time only helps with adequate information or model capability) and the choice of verification tools significantly shape agent behavior.